Why 3D is the Future of Feature Extraction

by Eric Andelin
28 June 2012

Tags3d, cardinal systems, lidar, modeling, point clouds, software

The creation of Ridley Scott’s Prometheus, released on June 8, began with the use of Google Earth and digital elevation maps to create a spectacular landscape in 3D. LiDAR plays an important role in the film itself as the characters scan a cave to see the topography in real time. This is the cutting edge of sci-fi VFX—but it also foreshadows a very real future for professionals who manage geospatial data.

Many of us already work with LiDAR, but we’ve been limited by the software in how we can manipulate the data. Feature extraction is typically handled in 2D cross-sections by manually identifying key points in the point cloud, such as the top, bottom and edge of a curb. Although automation has substantially streamlined this process, it still isn’t perfect. The reality is that we live in a 3D world and gather data in 3D. Shouldn’t we also be manipulating that data in 3D?

Software developers are starting to catch on. For example, Cardinal Systems’ VrLiDAR software allows users to display and edit LiDAR data in a true 3D point cloud with stereo glasses, much like photogrammetry but without the stereo pairs. The data can be viewed from any perspective and at any scale. Modeling roadways with curbs is simply a matter of exaggerating the elevation (Z) so that the breaks as well as the tops and bottoms of the curbs become visible. Other photogrammetry software developers, such as GeoCue and DAT/EM, have also developed innovative ways of working with LiDAR data directly in 3D. It’s a much simpler way to get through the data, and it allows service providers to offer a faster turnaround on deliverables.

There’s also another benefit. Many geospatial firms already have a few people on staff who understand stereo and 3D, and it’s much easier to teach those individuals to process LiDAR data than it is to expect a CAD technician to repeatedly cut cross sections. Using 3D-capable software allows us to use our existing teams to create innovative solutions with LiDAR data.

These tools don’t necessarily replace existing software. For now, there are still plenty of applications for which 2D is sufficient, and automated feature extraction will continue to be beneficial for users that can rely on preset parameters. But with 3D firmly embedded in our movies and our culture, it’s only a matter of time before our clients begin working with 3D data. When they do, our datasets should be ready to move into the 3D world.

Image: Woolpert

Eric Andelin, CP, GISP

Eric Andelin CP, GISP, in involved in geospatial business development for SAM Inc. He has 25 years of experience in mapping. He is passionate about the use of mobile LiDAR as an acquisition tool for high-accuracy mapping.


RELATED ARTICLES

To leave a comment on this story, please log in:

COMMENTS

Please login to comment

By posting a comment, you agree to the site's Terms & Conditions

On August 19, 2013 Jameson Harrington wrote:

The 3d classification in VR is is still the best I have found for cleaning up LiDAR data; however, it still remains the buggiest, crash prone software I have ever used, bar none.